Differential reproductive success of sympatric, naturally spawning hatchery and wild steelhead trout (Oncorhynchus mykiss) through the adult stage

Transcription

1 433 Differential reproductive success of sympatric, naturally spawning hatchery and wild steelhead trout (Oncorhynchus mykiss) through the adult stage Jennifer E. McLean, Paul Bentzen, and Thomas P. Quinn Abstract: We used multilocus microsatellite analysis to compare the reproductive success of naturally spawning wild steelhead trout (Oncorhynchus mykiss) with a newly established sympatric hatchery population in Forks Creek, Washington, U.S.A. Hatchery steelhead spawning in the wild had markedly lower reproductive success than native wild steelhead. Wild females that spawned in 1996 produced 9 times as many adult offspring per capita as did hatchery females that spawned in the wild. Wild females that spawned in 1997 produced 42 times as many adult offspring as hatchery females. The wild steelhead population more than met replacement requirements (approximately adult offspring were produced per female), but the hatchery steelhead were far below replacement requirements (<0.5 adults per female). The survival differential was greatest in the freshwater environment (i.e., production of seaward-migrating juveniles), but survival at sea favored the hatchery population in 1 year and the wild population in the next. The poor performance of the hatchery population may be a consequence of spawning too early in the winter, generations of inadvertent domestication selection, or a combination of these two. Résumé : Une analyse des microsatellites à plusieurs locus nous a permis de comparer le succès de la reproduction chez des truites arc-en-ciel anadromes (Oncorhynchus mykiss) sauvages qui frayent naturellement à celui d une population sympatrique nouvellement établie provenant d une pisciculture à Forks Creek, Washington, É.-U. Les truites de pisciculture qui frayent en nature ont un succès reproducteur nettement inférieur à celui des truites sauvages indigènes. Les femelles sauvages qui ont frayé en 1996 ont produit 9 fois plus de descendants adultes par femelle que les femelles de pisciculture qui se sont reproduites en nature. En 1997, les femelles sauvages ont produit 42 fois plus de descendants adultes que les femelles de pisciculture. La population sauvage de truite dépasse les taux nécessaires pour le remplacement de la population (environ 3,7 6,7 descendants adultes produits par femelle), mais les truites de pisciculture sont bien en deçà de ces taux (<0,5 adulte par femelle). La différence dans la survie est maximale dans les environnements d eau douce (i.e., la production des jeunes qui migrent vers la mer), mais la survie en mer a été plus grande une année chez les truites sauvages et l autre année chez les truites de pisciculture. Il se peut que la piètre performance de la population de pisciculture soit due à une fraye trop hâtive en hiver, ou à des générations de sélection involontaire de domestication, ou alors à une combinaison de ces deux facteurs. [Traduit par la Rédaction] McLean et al. 440 Introduction Variation in reproductive success among individuals of a species or population is the basis for evolutionary change, population differentiation, and local adaptation. Reproductive success is characterized by the number of offspring attributable to an individual or by the proportion of that individual s genes that persist into the next generation of breeding adults. Reproductive success thus incorporates the breeding success of the parents, including competition, mate choice and other forms of sexual selection, and the survival to breeding age of their progeny. Investigations of reproductive success must therefore examine processes that occur throughout the life history of the organism, as well as those that occur during its reproductive phase. Variation in reproductive success may arise from individual differences in morphology, behaviour, or life-history traits, as well as stochastic factors. Salmonids are well known for their homing behaviour, and reproductive isolation of populations is a consequence of homing (Quinn 1993). Repeated over many generations, isolation leads to evolutionary change as populations become adapted to conditions in their home rivers, such as temperature, flow regime, juvenile habitat, and spawning Received 26 February Accepted 2 April Published on the NRC Research Press Web site at on 29 May J17353 J.E. McLean, 1 P. Bentzen, 2 and T.P. Quinn. School of Aquatic and Fishery Sciences, University of Washington, Seattle, WA , U.S.A. 1 Corresponding author ( jenm34@washington.edu). Can. J. Fish. Aquat. Sci. 60: (2003) doi: /F03-040

2 434 Can. J. Fish. Aquat. Sci. Vol. 60, 2003 substrate (Taylor 1991; Quinn et al. 2001). Local adaptations include changes in characteristics associated with reproductive success such as adult body size, egg size, fecundity, and timing of reproduction. Through the process of homing and the resulting reproductive isolation of different populations, local adaptation creates an advantage for spawners of the home population. Adults returning to their natal stream will presumably be better adapted for the timing of high flows, substrate size, temperature, and other conditions in that stream than adults from other populations. Strays, or fish originating from and spawning in different streams, generally have lower reproductive success than fish returning to their natal stream (e.g., Mayama et al. 1989; reviewed in Quinn 1993). Salmon and steelhead hatcheries have been operating in North America and Europe for more than a century (Nielsen 1994) and concerns have been raised regarding their effects on wild salmon. Interactions between hatchery and wild fish are a significant issue in fishery management, in terms of both competition and interbreeding. Fish in hatcheries are subjected to different selective pressures than fish in the wild, and hatchery populations may become adapted to hatchery conditions, either through deliberate artificial selection or unintentional domestication selection. For example, steelhead trout (Oncorhynchus mykiss) of hatchery origin differed from their ancestral wild population in antipredator and agonistic behavior (Berejikian 1995; Berejikian et al. 1996). When hatcheries produce fish that spawn in the wild, these fish can be considered strays, and as such, may not attain the reproductive success of wild fish. Our long-term study at Forks Creek, Washington, U.S.A., contributes to a growing literature examining differences in fitness and reproductive success among wild hatchery and native nonnative populations, and the consequences of allowing their sympatric spawning (steelhead (Chilcote et al. 1986; Leider et al. 1990); coho salmon, Oncorhynchus kisutch (Fleming and Gross 1992); brown trout, Salmo trutta (Skaala et al. 1996; Cagigas et al. 1999; Hansen 2002); Atlantic salmon, Salmo salar (McGinnity et al. 1997; Crozier 2000; Fleming et al. 2000)). Here we build on our previous results concerning the number of seaward-migrating smolts produced per female from sympatric, naturally spawning hatchery and wild steelhead populations (McLean et al. 2003) by examining the differential fitness of these two groups through to the return of their adult offspring. Our objective was to classify adult offspring produced by these naturally spawning hatchery and wild fish to a population of origin, and compare the reproductive success of the hatchery and wild populations in terms of the number of adult offspring produced per female. Combined with our previous smoltproduction estimates, we estimated marine survival for both groups and assessed whether differences in performance between wild and hatchery populations resulted from differences in production in freshwater or marine environments. Materials and methods Study site Forks Creek is a tributary of the Willapa River in southwest Washington, and Forks Creek Hatchery is located approximately 250 m upstream of the confluence of Forks Creek and the Willapa River (Fig. 1). Prior to 1994, Forks Creek supported a small wild run of winter steelhead spawning from approximately March through May (Mackey et al. 2001). In 1994, Forks Creek Hatchery (which had produced Pacific salmon but not steelhead since 1895) received smolts from the Bogachiel Hatchery and began propagation of a hatchery steelhead run. The population from which these smolts were taken was derived from a combination of native Bogachiel River steelhead and the Chambers Creek stock, a generalized hatchery stock artificially selected for early spawning (November February; Ayerst 1977; Crawford 1979; Mackey et al. 2001). The hatchery smolts released into Forks Creek in this and all subsequent years were marked by the removal of their adipose fins. The first hatchery adults returned in the winter of and we designated them as brood year (BY) A weir across the creek prevents returning salmon and steelhead from migrating upstream of the hatchery and allows hatchery staff access to returning fish. Adult steelhead with intact adipose fins (i.e., naturally produced) were placed upstream. Those missing an adipose fin were taken for spawning in the hatchery. However, in the first 2 years when hatchery adults returned, after the hatchery s capacity for steelhead eggs was met, excess hatchery fish were allowed upstream to spawn in sympatry with the wild fish. This practice was then discontinued, and hatchery fish are no longer allowed upstream. Thus, the wild population was exposed to a discrete, 2-year pulse of hatchery influence. Sampling Our sampling of Forks Creek steelhead has been ongoing since 1996 (for a complete description of our sampling design see McLean et al. 2003). Here we examine adults allowed upstream of the weir to spawn in the wild in BY1996 and BY1997, and their adult offspring, sampled in the springs of The parents were of known origin (based on fin clips) and we classified the adult offspring as hatchery or wild, based on genetic analysis (see below). We then assigned them to BY1996 or BY1997, depending on their age. Scales, collected along with the fin clips and measurements of length and weight from each individual, were read to determine the age of the fish. The fish produced in the hatchery were all released after 1 year, and the freshwater age of fish produced in the wild was assumed to be 2 years if the scale was not readable, based on the unimodal size distribution of smolts, the presence of two juvenile size classes in Forks Creek in the fall, and general life-history pattern of steelhead in this part of their range (Busby et al. 1996). The number of years at sea was determined from scale examination. If the scale was unreadable, we used a length-at-age frequency distribution from known-age fish to assign an age. Each age assigned through this process received a confidence score calculated as the proportion of known-age fish of that type (male or female, hatchery or wild) at that size. Our genetic baseline data consisted of 55 wild and 362 hatchery adults that migrated upstream of the weir in 1996 and 1997, before any potential mixing of the hatchery and wild gene pools. Our unknown samples, those that we wished to assign to a population of origin, were the progeny of these two brood years. They had intact adipose fins, because their parents spawned in the wild. The progeny of BY1996 in-

3 McLean et al. 435 Fig. 1. Map of study-site location in Washington, U.S.A. The hatchery, the site of smolt and adult sampling, is located close to the confluence of Forks Creek and Willapa River ( ). The Willapa River flows northwest to the Pacific. The inset of Washington shows the location of Chambers Creek (1), the origin of the hatchery population, Bogachiel Hatchery (2), the origin of the hatchery smolts released into Forks Creek, and Forks Creek Hatchery (3). Weinberg equilibrium, tests of linkage disequilibrium, and genetic differentiation estimates between the hatchery and wild populations. GENECLASS (Cornuet et al. 1999) was used both to self-classify the adult hatchery and wild samples to determine accuracy and to assign the adult offspring of hatchery and wild steelhead that spawned naturally in the river to a parental population of origin. We used the Bayesian-likelihood-algorithm option and the as is procedure. A number of assignment tests were performed to evaluate the extent of separation between the hatchery and wild populations, to determine the best baseline dataset, and to estimate the likelihood of misclassification of offspring (see McLean et al. 2003). Briefly, to improve the accuracy of determination, we created a new baseline dataset by adjusting the sample size of the baseline hatchery population to match that of the baseline wild population. We did this by randomly selecting 55 hatchery adults from a pool of the 362 genotypes collected, and using these as the hatchery baseline samples. All 55 wild baseline samples were used. In the unknown sample assignment tests, we assigned all adult offspring to a population of origin. Reproductive success for each population was determined by dividing the number of adult offspring assigned to that population by the number of females of that type which had ascended the creek to spawn in the parental year. (In all but the most extreme divergence from a 1:1 sex ratio, the number of juvenile salmonids produced is determined by the number of females.) After offspring were assigned to a population, we examined differences in spawning date and size at maturity between the two groups of adult offspring. These numbers were compared between groups, as well as between parents and offspring. Through use of a cumulative-frequency histogram, we compared overall run timing of the hatchery and wild populations. Results cluded 2 fish at age 3 (i.e., returned in 1999), 74 fish at age 4 (in 2000), and 5 fish at age 5 (in 2001), for a total of 81 offspring. BY1997 did not produce any age-3 fish and 79 fish returned in 2001 as 4-year-olds. Genetic and data analysis Genomic DNA extractions and PCRs (polymerase chain reactions) of the eight microsatellite loci used were performed as outlined in McLean et al. (2003). PCR products were size-fractionated using a 96-well capillary system Molecular Dynamics MegaBACE 1000 (Amersham Biosciences UK Limited, Buckinghamshire, U.K.). Electropherograms were analyzed using Genetic Profiler software version 1.1 (Molecular Dynamics, Sunnyvale, Calif.). The GENEPOP (version 3.0) software package (Raymond and Rousset 1995) was used to calculate heterozygosity, probability tests of Hardy Population genetics Heterozygosities (H E ) ranged from 82.6 to 94.0% (average 88.7%) in the hatchery parent population and from 85.5 to 94.9% (average 90.5%) in the wild parent population. In the progeny of BY1996, H E ranged from 79.1 to 93.5% and averaged 88.5%. In the progeny of BY1997, H E ranged from 83.3 to 93.4% and averaged 89.2% (Table 1). Neither the progeny of BY1996 nor that of BY1997 had allele frequencies consistent with Hardy Weinberg proportions (P 1996 < , P 1997 = ). Both years expressed a heterozygote deficiency. There was no statistical support for linkage disequilibrium among any locus pair in either group of offspring. F ST values describing genetic differentiation between hatchery and wild populations differed among years and between parental and offspring groups. For the baseline dataset, the F ST value was For the offspring generations, values were F STProg1996 = and F STProg1997 = Population assignment Before BY1999, all adults with clipped adipose fins were of hatchery origin and most returned early in the season (November March), whereas those with intact adipose fins

4 436 Can. J. Fish. Aquat. Sci. Vol. 60, 2003 Table 1. Genetic variation of the eight microsatellite loci examined in the parent and offspring groups of steelhead trout, Oncorhynchus mykiss. BY1996 BY1997 S1998 S1999 AO1996 AO1997 Locus H E N A H E N A H E N A H E N A H E N A H E N A Oki Omy Omy1001UW Omy1011UW Omy1191UW Omy1212UW One Ssa Average Note: Data are shown for parents (BY1996 and BY1997) and offspring (smolts 1998 (S1998), smolts 1999 (S1999), adult offspring of BY1996 (AO1996), and adult offspring of BY1997 (AO1997)). H E is the expected heterozygosity (%) for each population. N A is the observed number of alleles. Table 2. Number of adult steelhead returning to Forks Creek over the course of the study. BY1996 BY1999 BY2000 BY2001 Hatchery Wild Hatchery Wild Hatchery Wild November December January February March April May June July Winter totals Spring totals Note: Before the first offspring of naturally spawning hatchery fish returned (2 in BY1999, but the majority in BY2000), most fish with intact adipose fins (produced in the wild) returned late. In BY2000 and BY2001, the 2 years when the vast majority of the offspring-of-interest returned, different return times were exhibited by offspring with intact adipose fins. were of wild origin and returned later in the season (April through July; see Mackey et al. (2001) and McLean et al. (2003)). In 1999 and later years, when the offspring of these hatchery and wild fish that spawned in the wild returned to Forks Creek as adults, there were 4 potential categories of fish entering the creek: (1) early clipped, (2) early unclipped, (3) late clipped, and (4) late unclipped. Of the 700 adults that returned early in the first 4 years (BY1996 BY1999), 98% had clipped adipose fins and were hatchery fish (Table 2). Of the 142 adults that returned late in those 4 years, almost 95% had unclipped adipose fins and were wild fish (Table 2). Categories 2 and 3 were almost absent, which is consistent with two separate hatchery and wild runs with different return times. When the progeny of BY1996 and BY1997 began to return, however, individuals with intact adipose fins could not be assumed to have wild parents. The percentage of early, unclipped fish increased more than fourfold. Because both hatchery and wild fish spawned in the river, it is possible that these naturally spawned adults had two hatchery parents, two wild parents, or one of each. We used the genetic data to determine their likely population of origin. Baseline-population data were as described in McLean et al. (2003). Overall, the level of correct self-assignment of the baseline data was 92%, and the log-likelihood-ratio (LLR) threshold criterion for zero incorrect assignments was ±0.8 (the LLR is the ratio of the log-likelihood value for assignment into one population versus the log-likelihood of assignment into the other). There was a bias in incorrect assignments; fish were more often incorrectly assigned to the hatchery population than the wild population, potentially inflating estimates of the reproductive success of hatchery steelhead in the wild. Adult progeny of 1996 spawners were assigned using the created baseline: 37 were assigned to the hatchery population and 41 to the wild population with a LLR threshold of zero. With a LLR threshold of ±0.8, we were unsure of the correct assignment of and therefore removed 11 hatcheryassigned adults and 13 wild-assigned adults. After this correction, 26 were assigned to the hatchery population and 28 to the wild population. Adult progeny of 1997 spawners

5 McLean et al. 437 Fig. 2. Distribution over day of return of log likelihood ratios (LLR) for offspring of (a) BY1996 and (b) BY1997. An LLR score of less than zero classified an individual as hatchery, and a score higher than zero classified an individual as wild. Day 0 is 20 November, day 115 is 15 March, and day 225 is 4 June. were assigned using the same created baseline dataset: 12 were assigned to the hatchery population and 67 were assigned to the wild population with a LLR threshold of zero. With a LLR cutoff of ±0.8, 7 individuals were removed from the hatchery population and 11 individuals were removed from the wild population, leaving 5 hatchery and 56 wild individuals. Therefore, the 90 hatchery females and 11 wild females from BY1996 produced hatchery adults and wild adults. The 73 hatchery females and 10 wild females in BY1997 produced 5 23 hatchery and wild adults. Reproductive success and marine survival Because the overall pattern is the same regardless of the estimate used (fewer hatchery adults produced than wild adults despite many times more hatchery females than wild females), hereinafter we use the LLR threshold of zero estimates (i.e., all returning adults, regardless of the certainty of their classification). Hatchery females spawning in 1996 produced an average of 0.41 adults per capita and hatchery females spawning in 1997 produced an average of 0.16 adults per capita. Wild females produced an average of 3.73 adults per capita in 1996 and 6.70 in With these estimates, and the smolt-production estimates from McLean et al. (2003), we calculated the marine survival of hatchery and wild steelhead. In 1996, hatchery females produced smolts with an overall marine survival of 38% and wild females produced smolts with a marine survival rate of 15% (Table 3). In 1997, hatchery smolts produced in the wild had a marine survival of 12% compared with 36% for the wild smolts (Table 3). Reproductive timing Wild and hatchery adult parents overlapped in migration date, but wild fish tended to arrive later (McLean et al. 2003), and their adult offspring followed a similar return timing pattern. We examined the relationship of the LLR assigned to an individual and that individual s return timing to determine whether certainty of population assignment was associated with this timing. Any hybrid offspring of hatchery and wild parents would presumably have an intermediate Table 3. Reproductive success of hatchery- and wild-raised steelhead spawning in the wild. Population LLR (close to 0) and may have made up the early, unclipped group of fish or exhibited an intermediate return timing. No association between timing and LLR was apparent (Fig. 2; Student s t test of mean day of individuals with absolute LLR values equal to or less than 0.8 versus individuals with values above 0.8, P 1996offspring = 0.137, P 1997offspring = 0.198). Figure 3 is a cumulative-frequency distribution showing the return timing (date passing the weir) of adults ascending Forks Creek. The origin (hatchery or wild) of parents was known from their adipose fin, and the offspring were all born in the creek (intact adipose fin) and assigned with genetic data to a population of origin. BY1996 parents and their offspring had similar return timing patterns (hatchery early, wild late), but the timing differed more between BY1997 and their offspring (more hatchery-assigned fish returned late) (Fig. 3). Discussion Female Smolts produced/female Marine survival (%) BY1996 Hatchery Wild BY1997 Hatchery Wild Note: Female refers to the number of females allowed access to the creek in that brood year. Smolts produced is the number of smolts produced, on average, per female (from McLean et al. 2003), marine survival is the estimated marine survival from smolt to returning adult stage, and RS is the overall reproductive success in terms of the number of adults produced, on average, per female parent. For the populations to meet replacement requirements, RS should be 2 or greater. Reproductive success and marine survival Hatchery steelhead spawning in the wild had lower reproductive success than native wild steelhead. Wild females RS

6 438 Can. J. Fish. Aquat. Sci. Vol. 60, 2003 Fig. 3. Return timing of parents and offspring. Days along the x axis range from 20 November (0) to 4 June (224), and the cumulative frequency of returns is presented on the y axis: (a) BY1996 parents; (b) offspring of BY1996; (c) BY1997 parents; (d) offspring of Broken lines represent hatchery returns and solid lines represent wild returns. produced 9 times (in 1996) and 42 times (in 1997) more adult offspring per capita than hatchery females spawning in the wild. The wild steelhead population more than met replacement requirements (>2 returning adults per female spawner), but hatchery steelhead spawning in the wild did not approach replacement requirements (<0.5 adults per female spawner). These results are consistent with our previous estimates of smolt production for these two groups of steelhead (McLean et al. 2003). A number of other investigations have also shown inferior performance of hatchery salmonids spawning in the wild compared with the performance of wild fish (McGinnity et al. 1997; Fleming et al. 2000; Hansen 2002). Thus, we must consider a question central to salmonid conservation: why do hatchery fish do so poorly when released into the wild? The answer may be different in different situations, and it may depend on species, population, location, the history of hatchery supplementation in the area, and broodstock collection protocol among other factors. Differential reproductive success in Forks Creek could be due to a number of differences between the populations; however, the most likely reasons here include the non-native origin of the hatchery fish, the altered timing of reproduction, and further domestication selection in the hatchery (for a more thorough discussion of these factors see McLean et al. 2003). The differential reproductive success of naturally spawning hatchery and wild steelhead was extensively examined on the Kalama River (Chilcote et al. 1986; Leider et al. 1986, 1990). There, naturally spawning, wild, summer steelhead produced times as many adult offspring as did sympatrically spawning, hatchery steelhead (Leider et al. 1990), which is consistent with the reproductive success patterns that we observed. The Kalama researchers sampled at different life-history stages to determine when differential mortality between hatchery and wild offspring was the greatest, and determined that hatchery survival was particularly (and equally) poor at the subyearling-to-smolt stage and the marine phase. Although our sampling prevented us from examining the survival from subyearling-to-smolt stages, we did estimate the marine survival of the two groups. Markedly different results were obtained from the two brood years examined. Marine survival for the two groups in 1996 was 38% for the hatchery fish and 15% for the wild fish, and in 1997 the opposite occurred, i.e., 12% for the hatchery fish and 38% for the wild fish. Taken together, there was no clear differential between wild and hatchery fish, in contrast to the finding from the Kalama River work. It was hypothesized that wild fish would have a large advantage over hatchery fish in fresh water because these two types of fish experience radically different environments during all life stages in fresh water (spawning, incubation of embryos, and growth of juveniles). However, both types experience similar regimes of natural selection at sea, so their survival rates at sea were predicted to be similar. In salmonids, survival at sea is often influenced by the size of smolts, the timing of their entry into the ocean, or a combi-

7 McLean et al. 439 nation of these factors (e.g., Ward and Slaney 1988; Henderson et al. 1995; Salminen and Kuikka 1995). Neither of these factors, however, explained the differences in survival we observed because the wild and hatchery smolts were similar in both timing and size at emigration (McLean et al. 2003). It is not clear which (if either) year best represented the marine-survival rates of the two populations, so we are reluctant to overinterpret the patterns. The parsimonious conclusion that there was no difference between forms should stand until further data are available. Reproductive timing and hybridization The timing of reproduction is highly heritable in salmonids (rainbow trout, Salmo gairdneri (Gall et al. 1988; Siitonen and Gall 1989; Su et al. 1999); chinook salmon, Oncorhynchus tshawytscha (Quinn et al. 2000); pink salmon, Oncorhynchus gorbuscha (Smoker et al. 1998)). Thus, the reproductive timing of the assigned offspring compared with the timing of the known parents should give insight into whether hybridization has occurred. A cumulative-frequency distribution of adults passing the weir shows the distinct return timing patterns of the two groups before any chance of hybridization. In 1996, 90% of the hatchery steelhead had returned by March, but fewer than half of the wild adults had returned by this time; the majority of the wild adults returned in May. In 1997, the hatchery fish again returned early and wild fish returned late. In the next generation, the patterns changed slightly, and there were more late-returning hatcheryassigned fish (especially for the progeny of BY1997). The blending of return times between the two groups, especially for the progeny of 1997, is consistent with the possibility of interbreeding and hybridization. Because of the number of shared alleles, overlap in allele frequencies, and low F ST value between the two populations, it is not possible in this analysis to determine if the overlap in return timing of the parental generation resulted in any hybrid offspring between hatchery and wild fish. However, in addition to the altered return timing of the offspring generation relative to the timing of pure parental forms, there is indirect genetic evidence for some hybridization. There were many more offspring than parents with likelihood ratios close to zero (26 vs. 8%). The ratios are negative for the hatchery population and positive for the wild population, and so would be close to zero for hybrids. Hybridization has been documented among native and non-native or wild and hatchery populations in a number of species (McGinnity et al. 1997; Crozier 2000; Hansen 2002), and may be occurring in Forks Creek steelhead as well. Although currently the question of hybridization remains unanswered, future analysis of specific parentage may enable us to determine the extent of interbreeding between the two groups, and the survival rates and reproductive success of their hybrids. With the poor reproductive success of the hatchery fish, hybridization between the two groups will likely have a negative effect on the wild population, and potentially decrease the reproductive success of the wild group. If the productivity of the hybrids is intermediate between that of the two groups, the wild population faces potential loss of unique locally adapted gene complexes, a severe reduction in abundance, or even extirpation unless the hatchery genotype is quickly culled from the population. The cessation of releases of hatchery fish after the first 2 years provides the wild population with the opportunity to resist introgression. In conclusion, hatchery fish originating from a distant location and artificially selected for early return and spawn timing (and probably adapted to hatchery conditions) successfully reproduced in the wild. Although they produced offspring that survived to return to spawn themselves, the per-capita reproductive success of hatchery fish spawning in the wild was much less than that of the wild fish. The potential for hybridization between the two groups because of their overlap in timing may represent the most significant problem facing the wild steelhead population in Forks Creek. The hatchery group did not replace itself, and the survival rate and reproductive success of the hybrids are not yet known. Even with minimal differentiation at neutral loci, the difference in fitness between these two groups is significant. Salmonid conservation depends on knowing the consequences of introductions and artificial propagation on native populations, and our project contributes to the emerging picture provided by similar studies. Future analysis of this situation at Forks Creek will include specific parentage analysis to determine factors important for individual reproductive success such as body size, return timing and population of origin, and the fate of hybrids. The different results (relative marine survival of the two groups, changes in F ST and return timing) for the 2 years may be due to environmental differences between years or to differences in individual reproductive success and characters associated with freshwater and marine survival. Our long-term study of reproductive success at Forks Creek will afford us a rare view of what happens when hatchery fish are released into a wild population for the first time and their effect on subsequent generations. Acknowledgements We thank the many people involved in sample collection, especially the Forks Creek Hatchery staff and Greg Mackey. Long Live the Kings assisted with the trapping operation for the first 5 years of the project, and we particularly thank Brodie Smith and Larry Sienko for help in that regard. Our manuscript was improved by the valuable comments of two anonymous reviewers. This work was made possible by financial support from the Weyerhaeuser Company Foundation, the National Science Foundation (grant DEB ), and the Hatchery Science Reform Group in Washington State. References Ayerst, J.D The role of hatcheries in rebuilding steelhead runs of the Columbia River system. Am. Fish. Soc. Spec. Publ. No. 10. pp Berejikian, B.A The effects of hatchery and wild ancestry and experience on the relative ability of steelhead trout fry (Oncorhynchus mykiss) to avoid a benthic predator. Can. J. Fish. Aquat. Sci. 52: Berejikian, B.A., Mathews, S.B., and Quinn, T.P Effects of hatchery and wild ancestry and rearing environments on the development of agonistic behavior in steelhead trout (Oncorhynchus mykiss) fry. Can. J. Fish. Aquat. Sci. 53:

8 440 Can. J. Fish. Aquat. Sci. Vol. 60, 2003 Busby, P.J., Wainwright, T.C., Bryant, G.J., Lierhamer, L., Waples, R.S., Waknitz, F.W., and Lagomarsino, I.V Status review of west coast steelhead from Washington, Idaho, Oregon and California. NOAA Tech. Memo NMFS-NWFSC-27. Cagigas, M.E., Vazquez, E., Blanco, G., and Sanchez, J.A Genetic effects of introduced hatchery stocks on indigenous brown trout (Salmo trutta L.) populations in Spain. Ecol. Freshw. Fish, 8: Chilcote, M.W., Leider, S.A., and Loch, J.J Differential reproductive success of hatchery and wild summer-run steelhead under natural conditions. Trans. Am. Fish. Soc. 115: Cornuet, J.M., Piry, S., Luikart, G., Estoup, A., and Solignac, M New methods employing multilocus genotypes to select or exclude populations as origins of individuals. Genetics, 153: Crawford, B.A The origin and history of the trout brood stocks of the Washington Department of Game. Fisheries Research Report, Washington State Game Department, Olympia. Crozier, W.W Escaped farmed salmon, Salmo salar L., in the Glenarm River, Northern Ireland: genetic status of the wild population 7 years on. Rev. Fish Biol. Fish. 7: Fleming, I.A., and Gross, M.R Reproductive behaviour of hatchery and wild coho salmon (Oncorhynchus kisutch): does it differ? Aquaculture, 103: Fleming, I.A., Hindar, K., Mjolnerod, I.B., Jonsson, B., Balstad, T., and Lamberg, A Lifetime reproductive success and interactions of farm salmon invading a native population. Proc. R. Soc. Lond. B Biol. Sci. 267: Gall, G.A.E., Baltadano, J., and Huang, N Heritability of age at spawning for rainbow trout. Aquaculture, 68: Hansen, M.M Estimating the long-term effects of stocking domesticated trout into wild brown trout (Salmo trutta) populations: an approach using microsatellite DNA analysis of historical and contemporary samples. Mol. Ecol. 11: Henderson, M.A., Diewert, R.E., Stockner, J.G., and Levy, D.A Effect of water temperature on emigration timing and size of Fraser River pink salmon (Oncorhynchus gorbuscha) fry: implications for marine survival. In Climate change and northern fish populations. Edited by R.J. Beamish. National Research Council of Canada, Ottawa, Ontario. pp Leider, S.A., Chilcote, M.W., and Loch, J.J Comparative life history characteristics of hatchery and wild steelhead trout (Salmo gairdneri) of summer and winter races in the Kalama River, Washington. Can. J. Fish. Aquat. Sci. 43: Leider, S.A., Hulett, P.L., Loch J.J, and Chilcote, M.W Electrophoretic comparison of the reproductive success of naturally spawning transplanted and wild steelhead trout through the returning adult stage. Aquaculture, 88: Mackey, G., McLean, J.E., and Quinn, T.P Comparisons of run timing, spatial distribution, and length of wild and newlyestablished hatchery populations of steelhead in Forks Creek, Washington. N. Am. J. Fish. Manag. 21: Mayama, H., Nomura, T., and Ohkuma, K Reciprocal transplantation experiment of masu salmon (Oncorhynchus masou) population. 2. Comparison of seaward migrations and adult returns of local stock and transplanted stock of masu salmon. [In Japanese, with an English abstract.] Sci. Rep. Hokkaido Salmon Hatchery, 43: McGinnity, P., Stone, C., Taggart, J.B., Cooke, D., Cotter, D., Hynes, R., McCamley, C., Cross, T., and Ferguson, A Genetic impact of escaped farmed Atlantic salmon (Salmo salar L.) on native populations: use of DNA profiling to assess freshwater performance of wild, farmed, and hybrid progeny in a natural river environment. ICES J. Mar. Sci. 54: McLean, J.E., Bentzen, P., and Quinn, T.P Differential reproductive success of sympatric, naturally spawning hatchery and wild steelhead (Oncorhynchus mykiss). Environ. Biol. Fishes. In press. Nielsen, J.L Invasive cohorts: impacts of hatchery-reared coho salmon on the trophic, developmental, and genetic ecology of wild stocks, In Theory and application in fish feeding ecology. Edited by K.L. Fresh, D.J. Stouder, and R.J. Feller. University of South Carolina Press, Columbia. pp Quinn, T.P A review of homing and straying in hatcheryproduced salmon. Fish. Res. 18: Quinn, T.P., Unwin, M.J., and Kinnison, M.T Evolution of temporal isolation in the wild: genetic divergence in timing of migration and breeding by introduced chinook salmon populations. Evolution, 54: Quinn, T.P., Kinnison, M.T., and Unwin, M.J Evolution of chinook salmon (Oncorhynchus tshawytscha) populations in New Zealand: pattern, rate, and process. Genetica, 112/113: Raymond, M., and Rousset, F GENEPOP: a population genetic software for exact tests and ecumenicism. J. Hered. 86: Salminen, M., and Kuikka, S Annual variation in survival of sea-run Baltic salmon, Salmo salar L.: significance of smolt size and marine conditions. Fish. Manag. Ecol. 2: Siitonen, L., and Gall, G.A.E Response to selection for early spawn date in rainbow trout, Salmo gairdneri. Aquaculture, 78: Skaala, O., Jorstad, K.E., and Borgstrom, R Genetic impact on two wild brown trout (Salmo trutta) populations after release of non-indigenous hatchery spawners. Can. J. Fish. Aquat. Sci. 53: Smoker, W.W., Gharrett, A.J., and Stekoll, M.S Genetic variation of return data in a population of pink salmon: a consequence of fluctuating environment and dispersive selection? Alaska. Fish. Res. Bull. 5: Su, G.S., Liljedahl, L.E., and Gall, G.A.E Estimates of phenotypic and genetic parameters for within-season date and age at spawning of female rainbow trout. Aquaculture, 171: Taylor, E.B A review of local adaptation in Salmonidae, with particular reference to Pacific and Atlantic salmon. Aquaculture, 98: Ward, B.R., and Slaney, P.A Life history and smolt-to-adult survival of Keogh River steelhead trout (Salmo gairdneri) and the relationship to smolt size. Can. J. Fish. Aquat. Sci. 45: